API Reference

Base URL: https://dish-embed.latimal.com

Authentication

All endpoints except /health require an API key in the X-API-Key header:

curl -X POST https://dish-embed.latimal.com/embed \
  -H "X-API-Key: YOUR_KEY" \
  -H "Content-Type: application/json" \
  -d '{"items": ["Chicken Biryani"]}'

Keys are issued per-customer. Contact us to get yours.

Access is sold as monthly plans. One HTTP request counts as one API call against your plan's monthly quota, no matter how many items it carries. Plans also set a per-second rate limit and gate which endpoints are available:

Plan	API calls/mo	Rate limit	Endpoints
Starter	10,000	20 req/sec	/search, /classify, /match
Pro	100,000	60 req/sec	+ /suggest, /report
Scale	1,000,000	150 req/sec	+ /dedup

The Endpoints column lists what each tier adds over the one below it. The utility endpoints /embed, /embed/batch, /health, and /balance are available on every plan.

Every plan starts with a 14-day free trial. Reaching your monthly quota returns 402 with no overage billing. Bursting past your rate limit, or running too many requests at once, returns 429. See latimal.com/pricing for current rates.

Common patterns

All endpoints accept and return JSON
Per-item text limit: 500 characters
All text is automatically preprocessed (noise removal, spelling normalization) before processing
Embeddings are L2-normalized

Error codes

Code	Meaning
200	Success
400	Bad request (malformed JSON, invalid parameters)
401	Missing or invalid API key
402	Monthly quota reached (no overage billing; upgrade or wait for the next cycle)
422	Validation error (item too long, exceeds max count, etc.)
429	Rate limit exceeded (your plan's per-second limit, or too many simultaneous in-flight requests)
503	Model not loaded or server starting up

Latency (p50, production)

Endpoint	Items	Latency
POST /embed	100	~50ms
POST /embed/batch	1000	~400ms
POST /search	100 corpus	~80ms
POST /match	10 pairs	~120ms
POST /dedup	100 items	~200ms
POST /classify	100 items	~60ms
POST /report	500 items	~1.5s
POST /suggest	50 cart + 200 menu	~150ms

Endpoints

Endpoint	Description	Plans
POST /embed	Generate embeddings for menu items	All plans
POST /embed/batch	Batch embedding for large catalogs (up to 5K items)	All plans
POST /match	Check if pairs of items are the same dish	All plans
POST /dedup	Find duplicate items in a menu	Scale
POST /classify	Classify items by cuisine	All plans
POST /search	Semantic menu search	All plans
POST /report	Full menu health report	Pro, Scale
POST /suggest	Cart upsell suggestions	Pro, Scale
GET /health	Server status check	All plans
GET /balance	Check remaining quota	All plans